OPINION
DIXON W. HOLMAN, Justice. .Appellant Dionicio Vega Garza was charged with a non-capital felony offense. On the day set for trial, a venire was brought to the courtroom and seated so each venire-member could write answers to questions asked in a written questionnaire. Both the prosecutor and the defense attorney were present, observing the veniremembers while they completed their task. At 11:00 a.m., the questionnaires were answered, and the court immediately recessed for lunch. During the recess, both the prosecutor and the defense attorney reviewed the completed questionnaires in order to read the biographical and other information the veniremembers had provided about themselves. Court reconvened after lunch and, because of information seen in the veniremembers’ written answers, the State asked for a jury shuffle. See Tex. Code Crim. Proc. Ann. art. 35.11 (Vernon Supp.1999). Procédurally, both a defendant and the State have an equal opportunity to shuffle the venire. See id. The State’s request was granted over Appellant’s objection that relied upon and cited the trial court to the holding of Davis v. State, 782 S.W.2d 211, 214 (Tex.Crim.App.1989), cert. denied, 495 U.S. 940, 110 S.Ct. 2193, 109 L.Ed.2d 520 (1990). Complaints of jury shuffle error must be preserved for appellate review, and Appellant’s complaint about the shuffle was preserved when the trial court overruled his objection. See Johnson v. State, 977 S.W.2d 137, 139 (Tex.Crim.App.1998).
Appellant presents nine issues on appeal, but because the first issue is dispositive, we will not address the others. We will reverse and remand for a new trial.
The Issue
The issue is whether a party may evaluate the information gleaned from juror information cards or biographical questionnaires before deciding whether to request a shuffle of the venire. The Court of Criminal Appeals has said no. See Davis, 782 S.W.2d at 214. In both capital and non-capital offense cases, the Court of Criminal Appeals has stressed that a motion to shuffle the venire will not be granted unless timely made. Compare id. (a capital offense case) with Alexander v. State, 523 S.W.2d 720, 721-22 (Tex.Crim.App.1975) (a non-capital felony offense ease). In either type of case, a motion to shuffle is untimely and will be denied if the movant delays the motion until after the voir dire starts. See Davis, 782 S.W.2d at 214; Alexander, 523 S.W.2d at 721. In a non-capital case, the voir dire *354starts when the State begins examining the venire. See Williams v. State, 719 S.W.2d 573, 575 (Tex.Crim.App.1986). In a capital case, the voir dire starts when the trial judge begins examining the venire as prescribed in article 35.17 of the Texas Code of Criminal Procedure. See Davis, 782 S.W.2d at 215.
Here, Appellant argues that the State’s motion to shuffle should have been denied because it was made after voir dire had started in his non-capital felony case. The record reflects that when the motion was made, the State’s examination of venire-members on voir dire had not started. The State made its motion after the venire had been seated in the presence of counsel for Appellant and the State, after the venire-members had written their answers to the questionnaires, and after the parties’ lawyers had reviewed that written information during the lunch recess.
There is precedent for arguing, as the State does, that none of those events trigger the start of voir dire in a non-capital case. In 1975, the Court of Criminal Appeals established that the seating of a venire in a non-capital case does not start the voir dire. See Alexander, 523 S.W.2d at 721-22. In 1982, a court of appeals determined that the voir dire in a non-capital case does not begin when the attorneys review the juror information cards. See Holman v. State, 636 S.W.2d 18, 18-19 (Tex.App.—Dallas 1982, pet. ref'd). In 1986, the Court of Criminal Appeals held that in a non-capital ease, voir dire does not begin until “the State is recognized by the court to commence the voir dire examination and actually starts that examination.” Williams, 719 S.W.2d at 577. In 1987, the Court of Criminal Appeals determined that the voir dire does not begin in a non-capital ease just because the prosecutor tells the court, on the record, that he is ready to start the voir dire. See DeLeon v. State, 731 S.W.2d 948, 949 (Tex.Crim.App.1987). Then came the 1989 Court of Criminal Appeals holding in Davis.
Because of Davis, the determination of “timeliness” in connection with Appellant’s first issue involves more than the basic inquiry of whether or when the voir dire started. The crux of Appellant’s issue is whether a shuffle motion is untimely if it is made after the movant has read and evaluated the veniremembers’ personal information, furnished by them in writing at the court’s request. In addition to addressing when a voir dire starts, Davis emphasizes, making no distinction between capital and non-capital cases, that a party contemplating whether to ask for a shuffle of the venire under article 35.11 is only entitled to see its members before making the motion. Davis, 782 S.W.2d at 214.
For the purposes of a motion to shuffle, article 35.11 does not require a trial court to provide access to any personal information about the veniremembers. See id. (“We have never interpreted [article 35.11] as requiring the trial court to afford the defendant anything more than being able to view the outward appearance of the venire members.”). Based only on their outward appearance, a party may have one set of veniremembers discarded and another set summoned, using the mechanism of article 35.11. The article gives parties an ability to trade one proverbial “pig-in-a-poke” for another at a stage of the trial when neither party has the advantage of knowing any more about the potential jurors than their outward appearance. Alexander, decided by the Court of Criminal Appeals in 1975, noted that the Legislature did not intend for article 35.11 to enable parties to elect a venire shuffle based upon information elicited during the voir dire. Alexander, 523 S.W.2d at 721. The court also stated that “[t]o allow either party to request a shuffle of the names of the jury panel after voir dire begins would be disruptive and unduly prolong the trial.” Id. But, eleven years later, while adhering to the requirement that a motion to shuffle the venire must be made before the voir dire starts, the Court of Criminal Appeals also observed that “[t]he right of trial by jury stands orí a higher plane than expediency [and a] shuffle of the jury panel for the case actually takes a minimal amount of time if properly handled.” Williams, 719 S.W.2d at 577 n. 5 (emphasis added). In 1993, the Court of Criminal Appeals again scorned “expediency” as a measure in evaluating the propriety of a motion to shuffle the venire. *355Chappell v. State, 850 S.W.2d 508, 513 (Tex.Crim.App.1993) (“The right of trial by jury stands on a higher plane than expediency; and fair trial by jury means a jury selected according to the law regulating their selection and empanelment.”) (quoting Fontenot v. State, 379 S.W.2d 334, 335 (Tex.Crim.App.1964)).
When the venire shuffle issue in Davis came before the Court of Criminal Appeals, the court characterized it this way:
Appellant ... asserts that he was entitled to review the juror information cards and his questionnaires [biographical information] before he exercised his option to have the names shuffled. We disagree.
Davis, 782 S.W.2d at 213. Davis states plainly and simply why parties desiring a venire shuffle must ask for it based only on the outward appearance of the venire-members:
[I]t was not the intent of the Legislature [in article 35.11] to have the names of the venire shuffled based upon information obtained during voir dire....
See id at 214.
So too, Davis denies parties the use of information gleaned from juror information cards or biographical questionnaires as a basis for a motion to shuffle. Id. Significantly, in addressing the use of that source for information, Davis makes no distinction between capital and non-capital cases. Id. And, with respect to the use of juror cards and questionnaires as the basis for a shuffle, Davis makes a denial of the motion dependent upon the source of the information, the cards and questionnaires, not just dependent upon whether the voir dire has started. Id. Consequently, the timing of a motion to shuffle is still critical to the motion’s success, but for a trial court that is determining the viability of the motion, Davis makes pinpointing when the voir dire starts a consideration that is separate from the principle that forbids using the veniremembers’ written personal information as a basis for a motion to shuffle. Id.
Otherwise, Davis would stand for an untenable proposition: that although a venire shuffle must be premised only upon a view of the venire and therefore will be denied if based upon information verbally obtained during voir dire, the motion to shuffle will be granted if based upon identical information gleaned before voir dire from the venire-members’ own writings. That proposition is neither stated nor implied in Davis. Instead, a party who decides to make a shuffle motion either because of veniremembers’ verbal answers during voir dire or because of reviewing the personal information they have written about themselves, has taken the shuffle procedure well past the premise that a shuffle must be based merely upon the veniremembers’ outward appearance. See id.
Focusing on that rationale, the Court of Criminal Appeals held:
Following Alexander, [a non-capital offense case] we hold that because it was not the intent of the Legislature to have the names of the venire shuffled based upon information obtained during voir dire, so too it was not the intent of the Legislature [in article 35.11] to base the shuffle on information gleaned from juror information cards and/or biographical questionnaires.
Id. (emphasis added). Plainly, this holding is the antithesis of the court of appeals holding in Holman, seven years earlier. See Holman, 636 S.W.2d at 19.
In Appellant’s case, the prosecutor conceded that he relied on information gleaned from the questionnaires in deciding to ask for a shuffle. The lunch recess, during which the prosecutor reviewed the information the jurors had written, lasted more than an hour. Davis even addresses a significantly similar circumstance and rejects that length of time as incompatible with article 35.11:
Moreover ... counsel requested an hour to review the veniremen’s biographical information before he opted to shuffle their names. [Article 35.11] cannot be read to allow such....
Davis, 782 S.W.2d at 214.
Davis establishes that when a party is contemplating whether to request a jury shuffle, a trial court is not required to furnish any information about veniremembers be*356yond letting the party look at their outward appearance. Id. Davis also establishes the unmistakable principle that in a criminal court trial, regardless of whether for a capital offense or for a non-capital offense, the trial court is not permitted by article 35.11 to let the parties evaluate the information gleaned from juror information cards or biographical questionnaires before deciding whether to request a venire shuffle. Id.
Accordingly, to be timely, a motion to shuffle must be made before the voir dire begins, and also made before the movant has reviewed the juror information cards or biographical questionnaires. See id. This principle applies to the trials of non-capital as well as capital cases. We conclude that the trial court erred by granting the State’s motion to shuffle the venire.
Harm Analysis
Having found error, we will analyze its effect to determine whether the court’s judgment is reversible:
(a) Constitutional Error. If the appellate record in a criminal ease reveals constitutional error that is subject to harmless error review, the court of appeals must reverse a judgment of conviction or punishment unless the court determines beyond a reasonable doubt that the error did not contribute to the conviction or punishment.
(b) Other Errors. Any other error, defect, irregularity, or variance that does not affect substantial rights must be disregarded.
Tex.R.App. P. 44.2.
This court recently concluded that the rules governing jury shuffles have no constitutional origin and are purely a creation of the legislature. See Ford v. State, 977 S.W.2d 824, 826 (Tex.App.—Fort Worth 1998, pet. filed). Because the erroneous granting of a jury shuffle is not a denial of a constitutional right, we must determine whether it affects a substantial right.
The Court of Criminal Appeals recently said that a “substantial right” is “affected when the error ha[s] a substantial and injurious effect or influence in determining the jury’s verdict.” King v. State, 953 S.W.2d 266, 271 (Tex.Crim.App.1997) (citing Kotteakos v. United States, 328 U.S. 750, 776, 66 S.Ct. 1239, 1253, 90 L.Ed. 1557 (1946)); see Coggeshall v. State, 961 S.W.2d 639, 643 (Tex.App.—Fort Worth 1998, pet. ref'd). Applying the court’s definition from King to the context of Rule 44.2(b), the review standard for “other errors” is: any other error, defect, irregularity, or variance that does not have a substantial and injurious effect or influence in determining the jury’s verdict must be disregarded. See Tex.R.App. P. 44.2(b); King, 953 S.W.2d at 271.
In Ford, we relied on King’s definition of “substantial right” in conducting a “reversible-error” harm analysis of a trial court’s denial of a timely request for a jury shuffle. Ford, 977 S.W.2d at 827. We held that the term “substantial right” included certain statutory rights made mandatory by the Legislature, such as a right to a jury shuffle pursuant to article 35.11. Id. We concluded that it is impossible to determine whether an error that occurs before the jury is even empaneled affects that jury’s verdict and that “because we cannot measure whether [the jury shuffle] error had a substantial or injurious effect on the jury’s verdict, we cannot disregard the error as harmless.” Id. at 829. To do so would effectively make article 35.11 usable only at a trial court’s discretion.
The Court of Criminal Appeals recently addressed the application of the harm analysis to errors regarding jury shuffles. See Roberts v. State, 978 S.W.2d 580, 580 (Tex.Crim.App.1998). In Roberts, the trial court erroneously granted the State’s request for a jury shuffle after the voir dire ended. Id. In an unpublished opinion, the court of appeals had reversed Roberts’ conviction without conducting a harm analysis. Id. The Court of Criminal Appeals held that jury shuffle errors are not exempt from a harm analysis, explaining:
Except for certain federal constitutional errors labeled by the United States Supreme Court as ‘structural,’ no error ... is categorically immune to a harmless error analysis. Of course, where the error involved defies analysis by harmless error standards or the data is insufficient to *357conduct a meaningful harmless error analysis, then the error will not be proven harmless beyond a reasonable doubt under [former] Rule 81(b)(2) [now Rule 44.2].
Id. (quoting Cain v. State, 947 S.W.2d 262, 262 (Tex.Crim.App.1997)). The Court of Criminal Appeals noted that the court of appeals handed down its decision on August 29, 1997, two days before the new rules of appellate procedure took effect and before the Court’s opinion in Cain was handed down, and remanded the case back to the court of appeals to determine, in light of Cain and Rule 44.2(b), whether the jury shuffle error was amenable to a harm analysis, and if so, whether any harm occurred. Id.
In Cain, the trial court failed to admonish the defendant of the deportation consequences of a guilty plea. Cain, 947 S.W.2d at 263. The Court of Criminal Appeals applied the harm analysis provided in former Rule 81(b)(2), which mandated reversal of all criminal cases in which the record revealed an error “unless .the appellate court determine[d] beyond a reasonable doubt that the error made no contribution to the conviction or to the punishment.” Tex.R.App. P. 81(b)(2) (Vernon 1997, revised 1997). In applying former Rule 81(b)(2), the court explained that reversal is required unless the error can be shown harmless beyond a reasonable doubt and where the error is not subject to a harm analysis, it can never be proven harmless beyond a reasonable doubt. The Court concluded that “it may be true that some kinds of errors ... will never be harmless under the Rule 81(b)(2) test.” See Cain, 947 S.W.2d at 264.
The new rales of appellate procedure, which took effect on September 1,1997, created two separate harm analyses for errors in criminal cases. See Tex.R.App. P. 44.2. Under Rule 44.2(a), constitutional error must be reversed unless the error can be shown or determined to be harmless beyond a reasonable doubt. That rule creates a presumption that the error is harmful, and thus reversible, unless the presumption is rebutted. The logic of Cain is that if the error is not subject to a harm analysis, a presumption of harmful error prevails because the presumption cannot be rebutted. However, Rule 44.2(b) does not create any sort of presumption. Rule 44.2(b) merely requires the appellate court to ignore a non-constitutional error that does not affect a substantial right. A determination must be made as to whether a non-constitutional error does or does not affect a substantial right. Therefore, where a Rule 44.2(b) error does not lend itself to this type of a harm analysis, i.e., determining whether a substantial right was affected, neither logic nor Cain requires us to simply disregard the error.
A defendant’s right to a jury selected and empaneled according to the procedural rules and safeguards promulgated by the Legislature is a substantial right. See Chappell, 850 S.W.2d at 513 (“The right of trial by jury stands on a higher plane than expediency; and fair trial by jury means a jury selected according to the law regulating their selection and empanelmentl’)\ see also Williams, 719 S.W.2d at 577 n. 5. And, where the error violates a substantial right and occurs before the formation of the jury, the effect of the error is immeasurable. See Ford, 977 S.W.2d at 827 (“Determining whether an error had a substantial and injurious effect on a verdict is normally a difficult task for a reviewing court, but when, as here, the error concerns the formation of the jury itself, as opposed to error occurring in the presence of the jury, it is next to impossible to measure the harm from the fact of the record.”). To disregard a jury shuffle error under Rule 44.2(b) simply because an appellant is unable to show that the error affected the jury’s verdict, would leave to a trial court’s discretion the rule governing a jury shuffle. That would ignore the legislature’s prerogative to promulgate procedural rales for fairness in empaneling a jury. See Carranza v. State, 980 S.W.2d 653, 657 (Tex.Crim.App.1998). To allow an erroneous jury shuffle based upon information obtained from the voir dire’s juror information sheets “affects a substantial right” to a fair and random jury selection. Accordingly, we conclude that the error was harmful to Appellant and that the trial court’s judgment should be reversed. See Tex.R.App. P. 44.2(b).
*358Conclusion
We reverse the judgment and remand the case to the trial court for a new trial.
DAUPHINOT, J. filed a dissenting opinion.